[PATCH 0/3] net mshot fixes and improvements

[PATCH 0/3] net mshot fixes and improvements

[Pavel Begunkov]

The first patch fixes multishot interaction with io-wq for net
opcodes, followed by a couple of clean ups.

Pavel Begunkov (3):
  io_uring: fix mshot io-wq checks
  io_uring: refactor DEFER_TASKRUN multishot checks
  io_uring/net: dedup io_recv_finish req completion

 io_uring/io_uring.c | 20 ++++++++++++++++++++
 io_uring/net.c      | 37 ++++---------------------------------
 io_uring/rw.c       |  2 --
 3 files changed, 24 insertions(+), 35 deletions(-)

-- 
2.43.0

[PATCH 1/3] io_uring: fix mshot io-wq checks

[Pavel Begunkov]

When checking for concurrent CQE posting, we're not only interested in
requests running from the poll handler but also strayed requests ended
up in normal io-wq execution. We're disallowing multishots in general
from io-wq, not only when they came in a certain way.

Cc: stable@vger.kernel.org
Fixes: 17add5cea2bba ("io_uring: force multishot CQEs into task context")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 io_uring/net.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/io_uring/net.c b/io_uring/net.c
index 52f0d3b735fd..d4ab4bdaf845 100644
--- a/io_uring/net.c
+++ b/io_uring/net.c
@@ -87,7 +87,7 @@ static inline bool io_check_multishot(struct io_kiocb *req,
 	 * generic paths but multipoll may decide to post extra cqes.
 	 */
 	return !(issue_flags & IO_URING_F_IOWQ) ||
-		!(issue_flags & IO_URING_F_MULTISHOT) ||
+		!(req->flags & REQ_F_APOLL_MULTISHOT) ||
 		!req->ctx->task_complete;
 }
 
-- 
2.43.0

[PATCH 2/3] io_uring: refactor DEFER_TASKRUN multishot checks

[Pavel Begunkov]

We disallow DEFER_TASKRUN multishots from running by io-wq, which is
checked by individual opcodes in the issue path. We can consolidate all
it in io_wq_submit_work() at the same time moving the checks out of the
hot path.

Suggested-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 io_uring/io_uring.c | 20 ++++++++++++++++++++
 io_uring/net.c      | 21 ---------------------
 io_uring/rw.c       |  2 --
 3 files changed, 20 insertions(+), 23 deletions(-)

diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index cf2f514b7cc0..cf348c33f485 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -944,6 +944,8 @@ bool io_fill_cqe_req_aux(struct io_kiocb *req, bool defer, s32 res, u32 cflags)
 	u64 user_data = req->cqe.user_data;
 	struct io_uring_cqe *cqe;
 
+	lockdep_assert(!io_wq_current_is_worker());
+
 	if (!defer)
 		return __io_post_aux_cqe(ctx, user_data, res, cflags, false);
 
@@ -1968,6 +1970,24 @@ void io_wq_submit_work(struct io_wq_work *work)
 		goto fail;
 	}
 
+	/*
+	 * If DEFER_TASKRUN is set, it's only allowed to post CQEs from the
+	 * submitter task context. Final request completions are handed to the
+	 * right context, however this is not the case of auxiliary CQEs,
+	 * which is the main mean of operation for multishot requests.
+	 * Don't allow any multishot execution from io-wq. It's more restrictive
+	 * than necessary and also cleaner.
+	 */
+	if (req->flags & REQ_F_APOLL_MULTISHOT) {
+		err = -EBADFD;
+		if (!io_file_can_poll(req))
+			goto fail;
+		err = -ECANCELED;
+		if (io_arm_poll_handler(req, issue_flags) != IO_APOLL_OK)
+			goto fail;
+		return;
+	}
+
 	if (req->flags & REQ_F_FORCE_ASYNC) {
 		bool opcode_poll = def->pollin || def->pollout;
 
diff --git a/io_uring/net.c b/io_uring/net.c
index d4ab4bdaf845..14d6bae60747 100644
--- a/io_uring/net.c
+++ b/io_uring/net.c
@@ -78,19 +78,6 @@ struct io_sr_msg {
  */
 #define MULTISHOT_MAX_RETRY	32
 
-static inline bool io_check_multishot(struct io_kiocb *req,
-				      unsigned int issue_flags)
-{
-	/*
-	 * When ->locked_cq is set we only allow to post CQEs from the original
-	 * task context. Usual request completions will be handled in other
-	 * generic paths but multipoll may decide to post extra cqes.
-	 */
-	return !(issue_flags & IO_URING_F_IOWQ) ||
-		!(req->flags & REQ_F_APOLL_MULTISHOT) ||
-		!req->ctx->task_complete;
-}
-
 int io_shutdown_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 	struct io_shutdown *shutdown = io_kiocb_to_cmd(req, struct io_shutdown);
@@ -853,9 +840,6 @@ int io_recvmsg(struct io_kiocb *req, unsigned int issue_flags)
 	    (sr->flags & IORING_RECVSEND_POLL_FIRST))
 		return io_setup_async_msg(req, kmsg, issue_flags);
 
-	if (!io_check_multishot(req, issue_flags))
-		return io_setup_async_msg(req, kmsg, issue_flags);
-
 	flags = sr->msg_flags;
 	if (force_nonblock)
 		flags |= MSG_DONTWAIT;
@@ -951,9 +935,6 @@ int io_recv(struct io_kiocb *req, unsigned int issue_flags)
 	    (sr->flags & IORING_RECVSEND_POLL_FIRST))
 		return -EAGAIN;
 
-	if (!io_check_multishot(req, issue_flags))
-		return -EAGAIN;
-
 	sock = sock_from_file(req->file);
 	if (unlikely(!sock))
 		return -ENOTSOCK;
@@ -1403,8 +1384,6 @@ int io_accept(struct io_kiocb *req, unsigned int issue_flags)
 	struct file *file;
 	int ret, fd;
 
-	if (!io_check_multishot(req, issue_flags))
-		return -EAGAIN;
 retry:
 	if (!fixed) {
 		fd = __get_unused_fd_flags(accept->flags, accept->nofile);
diff --git a/io_uring/rw.c b/io_uring/rw.c
index 7733449271f2..6f465b6b5dde 100644
--- a/io_uring/rw.c
+++ b/io_uring/rw.c
@@ -933,8 +933,6 @@ int io_read_mshot(struct io_kiocb *req, unsigned int issue_flags)
 	 */
 	if (!io_file_can_poll(req))
 		return -EBADFD;
-	if (issue_flags & IO_URING_F_IOWQ)
-		return -EAGAIN;
 
 	ret = __io_read(req, issue_flags);
 
-- 
2.43.0

[PATCH 3/3] io_uring/net: dedup io_recv_finish req completion

[Pavel Begunkov]

There are two block in io_recv_finish() completing the request, which we
can combine and remove jumping.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 io_uring/net.c | 16 ++++------------
 1 file changed, 4 insertions(+), 12 deletions(-)

diff --git a/io_uring/net.c b/io_uring/net.c
index 14d6bae60747..96808f429b7a 100644
--- a/io_uring/net.c
+++ b/io_uring/net.c
@@ -685,20 +685,12 @@ static inline bool io_recv_finish(struct io_kiocb *req, int *ret,
 	if (msg->msg_inq && msg->msg_inq != -1)
 		cflags |= IORING_CQE_F_SOCK_NONEMPTY;
 
-	if (!(req->flags & REQ_F_APOLL_MULTISHOT)) {
-		io_req_set_res(req, *ret, cflags);
-		*ret = IOU_OK;
-		return true;
-	}
-
-	if (mshot_finished)
-		goto finish;
-
 	/*
 	 * Fill CQE for this receive and see if we should keep trying to
 	 * receive from this socket.
 	 */
-	if (io_fill_cqe_req_aux(req, issue_flags & IO_URING_F_COMPLETE_DEFER,
+	if ((req->flags & REQ_F_APOLL_MULTISHOT) && !mshot_finished &&
+	    io_fill_cqe_req_aux(req, issue_flags & IO_URING_F_COMPLETE_DEFER,
 				*ret, cflags | IORING_CQE_F_MORE)) {
 		struct io_sr_msg *sr = io_kiocb_to_cmd(req, struct io_sr_msg);
 		int mshot_retry_ret = IOU_ISSUE_SKIP_COMPLETE;
@@ -718,8 +710,8 @@ static inline bool io_recv_finish(struct io_kiocb *req, int *ret,
 			*ret = -EAGAIN;
 		return true;
 	}
-	/* Otherwise stop multishot but use the current result. */
-finish:
+
+	/* Finish the request / stop multishot. */
 	io_req_set_res(req, *ret, cflags);
 
 	if (issue_flags & IO_URING_F_MULTISHOT)
-- 
2.43.0

Re: [PATCH 0/3] net mshot fixes and improvements

[Jens Axboe]

On Fri, 08 Mar 2024 13:55:55 +0000, Pavel Begunkov wrote:
> The first patch fixes multishot interaction with io-wq for net
> opcodes, followed by a couple of clean ups.
> 
> Pavel Begunkov (3):
>   io_uring: fix mshot io-wq checks
>   io_uring: refactor DEFER_TASKRUN multishot checks
>   io_uring/net: dedup io_recv_finish req completion
> 
> [...]

Applied, thanks!

[1/3] io_uring: fix mshot io-wq checks
      commit: 3a96378e22cc46c7c49b5911f6c8631527a133a9
[2/3] io_uring: refactor DEFER_TASKRUN multishot checks
      commit: e0e4ab52d17096d96c21a6805ccd424b283c3c6d
[3/3] io_uring/net: dedup io_recv_finish req completion
      commit: 1af04699c59713a7693cc63d80b29152579e61c3

Best regards,
-- 
Jens Axboe